Content uploaded by Muhammad Adnan Khan
Author content
All content in this area was uploaded by Muhammad Adnan Khan on Apr 11, 2020
Content may be subject to copyright.
Available via license: CC BY-NC 4.0
Content may be subject to copyright.
International Journal of Computational Intelligence Systems
In Press, Uncorrected Proof
DOI: https://doi.org/10.2991/ijcis.d.200323.002; ISSN: 1875-6891; eISSN: 1875-6883
https://www.atlantis-press.com/journals/ijcis/
Regular paper
Blind Channel and Data Estimation Using Fuzzy Logic
Empowered Cognitive and Social Information-Based PSO
Muhammad Asadullah1,2, Muhammad Adnan Khan1,3,*, Sagheer Abbas1, Tahir Alyas3, Muhammad Asif Saleem3,
Areej Fatima1,3
1School of Computer Sciences, National College of Business Administration and Economics, E 40/ 1 Shahrah-e-Hazrat Imam Hussain, Block E 1 Gulberg III, Lahore,
54000 Lahore, Punjab, Pakistan
2Department of Computer Science & IT, The University of Lahore, 1 - Km Defence Road, 54000 Lahore, Punjab, Pakistan
3Department of Computer Science, Lahore Garrison University, Sector C, Phase VI, DHA, 54000 Lahore, Punjab, Pakistan
ARTICLE INFO
Article History
Received 28 Oct 2019
Accepted 13 Feb 2020
Keywords
MIMO
OFDMA
MC-CDMA
MUD
CSI
ABSTRACT
Multiple Input Multiple Output (MIMO) is a technology used to improve the channel capacity of the wireless communication
systems. Rapid increase in the number ofus ershas led to data rate demand increased in growing modern wireless communication
systems. To overcome this issue, MIMO is being used with several multicarrier techniques like Orthogonal Frequency Division
Multiple Access (OFDMA), Multi-Carrier Code Division Multiple Access (MC-CDMA), etc. Multi-user detection (MUD) with
artificial intelligence plays a vital role to enhance network capacity to meet the demands of future networks with an increased
number of users and multimedia services. Computational intelligence techniques are used in a multicarrier system to boost the
process of MUD. Some of the computational intelligencealgorithms like Swarm and Evolutionary are stuck in local minima and
due to this issue, the overall performance of the network decreases. For the convergence of Swarm intelligence-based solutions,
cognitive and social information (CSI) play a vital role. In this research article, the Fuzzy Logic empowered Cognitive and Social
Information (FLeCSI) algorithm using a fuzzy logic and swarm intelligence algorithm is proposed. By using social and cognitive
information FLeCSI updated each swarm position. After the simulation, it is observed that FLeCSI provides fast convergence
and minimize MMSE and BER as compared to techniques used previously for MUD like FLOMPSO, OLMPSO, Total Opposite
Mutant Particle Swarm Optimization (TOMPSO), Partial Opposite Mutant Particle Swarm Optimization (POMPSO), etc.
©2020 The Authors. Published by Atlantis Press SARL.
This is an open access article distributed under the CC BY-NC 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/).
1. INTRODUCTION
Due to rapidly growing wireless communication systems fast tech-
nological advancements are needed to meet the data transmission
requirements of developing communication systems. Instead of
physical wireline channels, wireless channels are used in the wire-
less communication system. Different types of services including
multimedia communication and voice to data are provided by the
wireless communication system. Rapidly growing wireless systems
needs rapid technological advancements to enhance the capacity of
current wireline systems. Undesired effects can be occurred due to
affected wireless signals because of physical properties and complex
interaction with the environment.
Due to signal scattering, around obstructing and large objects
the electromagnetic waves diffraction some problems that occur
between transmitter and receiver channel. Because of this prob-
lem signal effect with distortion, different attenuation, phase shift,
delays and inference of signal for multipath may be Beneficial or
detrimental. When destructive interface occurs the signal power
may be diminished slightly.
*Corresponding author: Email: madnankhan@lgu.edu.pk
For coherent or logically related detection of signal at receiving
antenna accurate channel state information (CSI) or channel esti-
mation (CE) is needed for most appropriate and enhancement in
performance of wireless communication system. If CSI is not avail-
able at the receiver, then differential demodulation technique is
used by non-coherent methods for detection and demodulation
of a signal transmitted. Non-coherent methods as compared with
the coherent detection method, due to the non-coherent detection
method loss in SNR cost about 3–4 dB loss. This situation directs
the research toward a coherent detection method at the receiver for
providing Channel Staten Information (CSI) in wireless communi-
cation systems.
For detection of inferring signals compressive sensing is used by
multi-user detection (MUD) as receiver technology. Sparse prop-
erty occurs in transmitting signal vector due to a large number of
non-zero elements in a situation in which mostly devices are not
inactive state and compressive signal (CS) problem occurs while
decoding of the transmitted signal at a receiver. Long term evolu-
tion is more suitable for a wireless communication system that has
a high activity a small number of users.
2M. Asadullah et al. / International Journal of Computational Intelligence Systems, in press
Research directs toward the enhancement of channel capacity with-
out affecting the quality of service for the modern wireless-based
networks for communication. For resolving the issues of channel
capacity and enhancing the data rates MIMO technique is more
effective [1–4]. In the MIMO method, the estimation of signals at
receiver and sender antennas is done [5] and increased the data rate
and bandwidth of the channel capacity [6–9].
To enhance the corresponding technique some antennas (transmit-
ter) and beneficiary radio are used. Depending on the amount of
data conveyed by the MIMO framework increment, the transmitted
information is calculated on several transmission paths [5]. While
on the receiving point, some receiver antennas receive information
to perform different calculations for the restoration of the informa-
tion to convert the data into the original form at the receiver [10].
For remote communication, MIMO technology is regarded as mid-
point and the amount of information without any additional data
transfer capacity and transmitting power because of an increase in
the scope and amount of information [11,12].
For enhancing the volume growth for several correspondences, the
Orthogonal Frequency Division Multiplexing (OFDM) and mul-
ticarrier code division multiple access (MC-CDMA) can also be
utilized with medium MIMO innovation technique [5–7,13]. One
of the optimal detectors is the maximum likelihood (ML) method
but is difficult to use for the achievement of exponential com-
plexity. Minimum mean square error (MMSE) detector/locator in
Machine to Machine (M2M) and the Maximum a Posteriori (MAP)
or marginal likelihood detectors, the sub-optimal MUD detectors
for example null steering or the zero forcing detector are used in
a less complex situation. The data sent simultaneously by multi-
ple servers to share multi-channel access on the bases of strategies
used for demodulation which is the main issue of the MUD. The
last two suboptimal techniques are very simple as they use matrix
inversion. Evolutionary algorithms like fuzzy adaptive differential
evolution (FADE), repeated weighted boosting search (RBS) and
differential evolution algorithms (DEAs) are very useful for MUD
and CE. Continues Search Space used for CE and discrete search
space for MUD. The Multiuser-MIMO (MU MIMO) broadcasting
techniques are most widely used for improvement of the spectral
efficiency. The quality of pre-coding transmitting to dominate the
multiuser inference degraded because of course knowledge of CSI
at the transmitter. Hence, there are chances that the interface from
co-scheduled user equipment will affect system throughput.
Therefore, the several new techniques like Particle Swarm
Optimization (PSO), Total Opposite Mutant Particle Swarm Opti-
mization (TOMPSO) [5,7,8,13], Partial Opposite Mutant Particle
Swarm Optimization (POMPSO), Island Differential Equation
(IDE), Genetic Algorithm (GA) and Island GA could be used
for enhancing the performance and capacity of modern digital
communication system [6,7,13–17].
In correspond for both antennas, i.e. sender and receiver, we
performed CE for high data rates in this research. Some distor-
tion added to the sending signal during transmission through
the strength of the weakens signal and the channel, due to this
scenario the receiver may not be able to get actual transmitted
information. For avoidance of said situation, fuzzy logic used to
enhance the channel and data estimation procedure [7,8]. In this
article Fuzzy Logic empowered Cognitive and Social Information
(FLeCSI)–based new variant of particle swarm intelligence tech-
nique used to fulfill/meet the requirements/demands of future
generation communication systems.
In this research work, the MIMO system is used because it consists
of a different number of users. It is assumed that the channel is cyclo
stationary and flat fading.
The core contributions of the article are enumerated as follows:
(a) Formulate an optimization problem for that the main
objective is to minimize MMSE, Minimum Mean Channel
Error(MMCE) and Bit Error Rate(BER) for optimum utiliza-
tion of bandwidth.
(b) For the better approximation of the user data and the chan-
nel coefficients, Fuzzy Logic empowered the Cognitive Social
Information (FLeCSI) algorithm with two variants FLeCSI-
TOLMPSO and FLeCSI-POLMPSO has been proposed.
(c) Proposed FLeCSI method compared with various state of
art algorithms like TOMPSO and POMPSO [13,18] and FL-
OLMPSO [18]. Simulation results have shown that FLeCSI
gives far better results as compared to other algorithms.
The sequence of the article is as follows: Literature review, MIMO
system model, the proposed FLeCSI-based optimization problem is
expressed in Sections 2,3, and 4. Section 5represents the simulation
results and discussion. Afterward, the research work conclusion and
future work in Sections 6and 7.
2. LITERATURE REVIEW
PSO algorithm has been proposed by [19,20] based on the observa-
tion of bird’s swarm in search of food. According to the PSO algo-
rithm each bird acts as a particle updating his position on the bases
of starting velocity, the last position visited personally, i.e. gath-
ered information about neighboring best position from social activ-
ities, i.e. social component and cognitive component. PSO provides
faster convergence in some optimization problems as compared to
the evolutionary algorithm [21], the second thing is the variation
of parameters is less. PSO is sensitive to fall into local optimum
points when optimizing complex multidimensional functions that
is its major drawback. Some qualities of PSO including greater and
easy implementation researchers attracted to work with PSO. Sev-
eral modifications have been done in the PSO algorithm that is
applied in different applications of fuzzy logic, neural network [22]
and many more.
For batter control over the velocity of swarm particles and to get
overall better performance Inertia factor (w) inducted in standard
PSO. On the basis of logistic human population growth, a new
Incremental Particle size PSO (IPPSO) has been proposed in which
the size increases with every iteration. For achieving an overall
improvement in the performance of PSO a fast convergence PSO
algorithm in which a new parameter named particle mean dimen-
sion has been proposed for the balance of diversity of each swarm
particle [23]. The variant of the PSO algorithm that ensures con-
vergence of the searching method with better solutions than stan-
dard PSO based on the construction factor approach [24]. Hybrid
PSO incorporating dynamic chaos improved the efficiency of the
M. Asadullah et al. / International Journal of Computational Intelligence Systems, in press 3
algorithm [25]. For stable convergence of complex multidimen-
sional problems, PSO particles ability have been explored by [24].
Exploration and exploitation capacity of GA and PSO have been
compared [26]. A novel hybrid algorithm based on PSO and GA
have been proposed [27]. Incorporated PSO with the technique of
tournament selection where current best particle properties trans-
ferred into inferior ones. Suganthan et al. [28] proposed an algo-
rithm named Local Best (LBest) model, in which the whole swarm
population divided into a small neighborhood and each neighbor-
hood maintain its LBest itself and is very well as in local optima
solution it avoids trapping of the swarm but its convergence rate
is very low. An adaptive inertia weight PSO algorithm has been
proposed by [29].
A new variant of PSO have been introduced for reaching global
optima reduce exploitation time and non-linear variation of iner-
tia weight for improvement of search capacity in multidimensional
space has been used [30]. By adding dynamic behavioral inertia
factor and mutating the particles of PSO adaptively PSO perfor-
mance has been improved [31]. An improved PSO technique pro-
posed by [32] used for the identification of parameters in chaotic
dynamical systems. Brits et al. [33] proposed the Niche PSO tech-
nique used for several global optimum solutions for multimodal
optimization. An evolutionary algorithm BSA [34] like PSO while
looking for global optima chooses the direction of an individual to
move from any particles of the last generation randomly. A number
of different unimodal optimization problems have been explored
and exploited by using the PSO algorithm [35]. To include the
effect of inertia term in PSO [36] studied the trajectories of general
swarm particles. A novel reformed modified version of PSO like
Cognitive and Social information-based Particle Swarm Optimiza-
tion (CSI-PSO) proposed by [37] in which instead of the velocity
concept of a swarm particle each swarm particle updates its posi-
tion from its own position in search space using its personal and
social information.
3. SYSTEM MODEL
Figure 1shows “n” no. of antennas at the receiver end. After
receiving signals frequency down is performed. Remove Cyclic Pre-
fix which adds at the transmitter end and then convert it into
Serial to Parallel (S/P). After S|P Fast Fourier Transform (FFT)
is performed.
There are A transmit antennas and O receiving antennas. The
channel implemented is a flat fading channel. During the commu-
Figure 1 Proposed hybrid computational intelligence approach-based
multiple input and multiple output receiver system model.
nication process of S symbols, the channel is expected to be station-
ary. The received signal at receive antenna o is as [1]:
ro(i)A
ano.aea(i)vo(i)(1)
where iis the index of symbol, no.ais the flat fading channel coeffi-
cient that links transfer antenna a to receiver antenna o, ea(i)is ith
symbol transmitted from antenna a taking value from the symbol
set {−1, +1} of Binary Phase Shift Key (BPSK) and vo(i)is Additive
White Gaussian Noise (AWGN) with F vo(i)
V.
The MIMO channel equation below will represent the complete
system:
r(i)Ne (i)v(i)(2)
where v(i)represents AWGN.
v(i)v(i)v(i)vO(i)T(3)
The transmitted symbol vector is
e(i)e(i)e(i)eA(i)T
and the received signal vector is
r(i)r(i)r(i)ro(i)T
The channel gain at receive antenna can always be normalized to
unity.
A
anoa
where, N(oa)no.a
Now defined a received data matrix with O × V dimensions and
transmitted data matrix with A*V dimensions as [1]:
Rr()r()r(S)(4)
and
Ee()e()e(S)(5)
respectively. Then the PDF of the received signal matrix R condi-
tioned on the MIMO channel matrix H and the transmitted data
matrix F can be written as
Prob R
NE
vOS e–
vS
ir(n)–Ne(i)(6)
The ML estimation of the transmitted symbols E and the MIMO
channel matrix H can be obtained by maximizing the Prob R
NE
over N and E mutually. Equally, the joint ML estimation can be
obtained by minimizing the following cost function:
JML
E
N
O x S
S
ir(i)–
Ne(i)(7)
Namely, the joint ML CDE is obtained as
JML
E
Narg min
T
NJML
E
N(8)
Equation (7) demonstrates that the search for the optimal joint ML
solution is over the discrete space of the transmitted symbols and
the continuous space of the MIMO channel matrix mutually.
4M. Asadullah et al. / International Journal of Computational Intelligence Systems, in press
3.1. Improved Cost Function
Equation (7) can be written as
JML
E
N
OSS
ir(i)–S
ir(i)
Ne(i)S
i
Ne(i)
The procedure above mentioned is included in any MUD and CE
for any multi-user MIMO system. MMSE-based batch processing
receiver, cost function and derivation is given below [18].
JML
E
N
OSS
ir2(i)–S
ir(i)
Ne(i)
S
i
Ne(i)
(9)
In the above equation, “S” Total symbols are transmitted over “O”
receive antennas. Then we let,
CML
E
N S
ir(i)
Ne(n)–
S
i
Ne(i)(10)
Putting the values from Equation (1) to (2)
JML
E
N
O x S S
ir(i)–CML
E
N(11)
Equation (1) can be written as
JML
E
Nmin
ENS
ir2(i)–CML
E
N(12)
It is observed that joined ML CDE can be presented as [18]
JML
E
Nmax
E
N
CML
E
N
4. PROPOSED FUZZY LOGIC EMPOWERED
CSI-BASED OPPOSITE SWARM
OPTIMIZATION ALGORITHM
Opposite Swarm Optimization used with fuzzy logic to enhance the
velocity with the help of fuzzy logic controller which takes two input
parameter as local and global intelligence and provide the cognition
and social information. Algorithm performed following different
steps for the preparation and evaluation of data.
Step 1: To start and initialize the populaces of data Pa= {Pa1, Pa2,
..................Pae} and velocity Vd
Step 2: To calculate cost of population utilization.
Step 3: To calculate the value of Lower and Upper Bound (LPb)
and (UPb) from Pbindependently.
Step 4: To calculate the opposite populace given as:
•Total Opposite Learning
Opposite Data Population
ϘPa= {ϘPa1, ϘPa2,..................ϘPae }
ϘPai = {ϘPai,1,ϘPai,2,.................. ϘPai,M }
ϘPai,j =Wƃp+ßƃa- Pai,j.
•Partial Opposite Learning
Opposite Data Population
ϘPa= {ϘPa1, ϘPa2,.................. ϘPae/2 }
ϘPai = {ϘPai,1,ϘPai,2,.................. ϘPai,M }
ϘPai,j =Wƃp+ßƃa- Pai,j.
Step 5: To calculate the appropriateness of Opposite Populations
(ϘPa)
Step 6: Choose the value of data population in Local best particle
ϻbpa from Paand ϘPaand Global best particle of Gbpa
from Paand ϘPa.
After performing the above-mentioned steps, select and update
the velocity of Global best particle for data population
Step 7: Calculate Local & Global Intelligence:
(LI) = Pij - LPbij (n−1) and (GI) = Pij - GPbij (n−1)
[FLCO, FLSO] = FLC (LI, GI)
Vdij(n) = Vdij(n - 1)FLCO (Local Intelligence) LI + FLSO
(Global intelligence)
Step 8: Calculate the data population particle of fitness and
update the value of Pa.
Step 9: The same iterations perform for the next received signal
until achieved the desired MSE or NoC.
Step 10: Finally Stop
The membership function of input and output variables used in
the proposed system mathematically and graphically are presented
below (Table 1).
4.1. Fuzzy Propositions
A fuzzy proposition represents the union, intersections and com-
plement by using the connectives “or,” “and” & “not.” Finally, the
fuzzy propositions can be written as
t:li gi FLCOFLSO(13)
Here, li,gi,FLCO and FLSO represent Local Intelligence, Global
Intelligence, Fuzzy Logic empowered Cognition and Fuzzy Logic
empowered Social information respectively. All values are repre-
sented in the range of 0 to 1 in both the I/O variable. Equation (14)
shows the t-norm function for the final layer as
t: (14)
Proposed fuzzy inference system along with membership function
defined as
tL(li)Ggimin L(li)Ggi(15)
Equation (15) can be written as
LGligitL(li)Ggi(16)
From Equations (15) and (16)
LGligiL(li)Ggi(17)
M. Asadullah et al. / International Journal of Computational Intelligence Systems, in press 5
Table 1 Membership function variables of input/output used in the proposed Fuzzy Logic empowered Cognitive and Social Information
(FLeCSI)-OSO-based system.
Sr # Input/Output
Variables
Representation of Membership
Functions (MF) Mathematically
Representation of Membership
Functions (MF) Graphically
1 LocalInt = LILi (l)
Lismall (l)
–l
l
LiMedium (l)
l
–l
–l
l
LiLarge (l)
l
–l–
)
2GlobalInt = GIGi g
Gismall g
–g
g
GiMedium g
g
–g
–g
g
GiLarge g
g
–g–
)
3 FLCO = FC µFc(fc)
Fclow (Fc)
.–fc
.–.
Fcmedium (Fc)
fc .
.–..–fc
.–.
Fchigh (Fc)
fc –.
.–.
4 FLSO = FS µFs(fs)
Fslow (Fs)
.–fs
.–.
Fsmedium (Fs)
fs .
.–..–fs
.–.
Fshigh (Fs)
fs –.
.–.
6M. Asadullah et al. / International Journal of Computational Intelligence Systems, in press
4.2. Lookup Table
Table 2shows the lookup table for a proposed system that contains
9 I/O fuzzy rules.
Nine rules of Fuzzy at the final layer denoted by Rcpµ, while (1 ≤ µ
≤ 9).
Rcp4=For Medium local intelligence & Small global intelligence
THE RESULT IS medium cognitive information and high
social information.
Rcp5=For Medium value of both L(li)&GgiTHE RESULT
IS also medium value of both (cognitive and social)
information.
Rcp6=For Medium L(li)& Large GgiTHE RESULT IS
medium cognitive information and low social information.
.
.
.
Rcp9=For Large value of both L(li)&GgiTHE RESULT IS
low cognitive information and low social information.
At the final layer nine Fuzzy rules denoted by Rcpµ, while (1 ≤
µ ≤ 9).
4.3. Inference Engine
The core part of Fuzzy inference is the fuzzy logic operators, mem-
bership functions and IF-THEN rules. Union operator is used for
combining the rules of fuzzy relations.
Proposed FLeCSI-based expert system IF-THEN rules be denoted
as Rȹof the final layer; which is,
RȹLGU
vcU
vc(18)
Equation (18) can be written as
LGligi(li)Ggi(19)
The single fuzzy relation using the rules of the final layer is
inferred as
R
Rȹ(20)
Table 2 Lookup table for proposed FLeCSI-OSO-based MC-CDMA
system.
Rules Local
Intelligence
Global
Intelligence
Cognitive
Information
Social
Information
1 Small Small High High
2 Small Medium High Medium
3 Small Large High Low
4 Medium Small Medium High
5 Medium Medium Medium Medium
6 Medium Large Medium Low
7 Large Small Low High
8 Large Medium Low Medium
9 Large Large Low Low
FLeCSI, Fuzzy Logic empowered Cognitive and Social Information; MC-CDMA, Multi-
Carrier Code Division Multiple Access.
Consider ith and Wbe fuzzy set and fuzzy inference engine of input
and output individually. We get the output fuzzy inference engine to
view IF-THEN rule R9by using the generalized modus ponens as
SlowMediumHigh WcWc
supi(LG)tiligiRligiUvcRligiUvc
(21)
The equation of Product Inference Engine (PIE) can be written as
FLCOFLSO
supi(LG)
jLGljgjiii(ƃƃƃ)
(22)
4.4. De-Fuzzifier
De-fuzzier of the proposed system specifies the by the member-
ship function of Ψ, i.e.
∫b
ad
∫b
ad ∫b
ad
∫b
ad
(23)
The de-fuzzifier of the proposed FLeCSI-based system is shown
below in Figures 2and 3.
Figure 2shown that if L(li)is small, medium or large for any value
of Ggithen cognitive information is also low, medium or high,
respectively.
Similarly, it also observed from Figure 3if Ggiis small, medium
or large for any value of L(li)then social information is also low,
medium or high, respectively.
4.5. Lookup Diagrams
Figures 4–7shows the lookup diagrams for Proposed Fuzzy Logic
empowered CSI-based Swarm Optimization with all possible cases
Figure 2 Proposed Fuzzy Logic empowered Cognitive and Social
Information(FLeCSI)-based rule surface for cognitive information of
local intelligence and the global intelligence.
M. Asadullah et al. / International Journal of Computational Intelligence Systems, in press 7
of the Updated Cognitive and Social Information. Figure 4shows
that if L(li)is small & Ggiis large the cognitive information is
high and social information is low.
Figure 5shows that if L(li)is large and Ggiis small the cogni-
tive information is low and social information is high.
Figure 6shows that if both L(li)and Ggiare medium then both
CSI are also medium.
Figure 7shows that if both L(li)&and Ggiare small then CSI
are also high.
Figure 3 Proposed Fuzzy Logic empowered Cognitive and Social
Information(FLeCSI)-based rule surface for social information of local
intelligence and the global intelligence.
Figure 4 Proposed Fuzzy Logic empowered Cognitive and Social
Information(FLeCSI)-based lookup diagram for high CI and low SI.
Figure 5 Proposed Fuzzy Logic empowered Cognitive and Social
Information (FLeCSI)-based lookup diagram for low CI and high SI.
4.6. Results and Discussions
For the implementation of the Blind Channel and Data Estima-
tion (BCDE) MIMO system, BPSK signaling scheme is used at
data sequence length Q = 50. All transmitters are equipped with
A = 3 to transmit antennas and O = 3 antenna used for the base
stations. Implementation of the Rayleigh flat fading channel with
four paths. The transmitter carrier frequency 900 Megahertz, which
corresponds to the Doppler frequency of 25 Hertz also moving at
the speed of 30 km/per hour. For simulation 3*k MIMO channel is
used having k = 10 users with a data population of 100 and 5 no. of
cycles for both algorithms. Channel population size is 5*Phwhere
Phrepresents the size of the channel matrix.
However, the simulation performance is measured by the MMCE,
which is:
MMCE
AO
A
a
O
onao–
N(ao)(24)
The performance and comparison of Proposed FLeCSI based
on both variants with previous algorithms are represented in
Tables 3and 4for channel and data estimation of MIMO sys-
tem with respect to Number of Cycles (NoCs) v/s MMSE, Miss
Rate/BER and MMCE, respectively.
Table 3shown that Conventional POMPSO and Conventional
TOMPSO proposed in [13,18] gives 10−3 and 10−5.4 MMSE at 160th
and 180th NoCs, respectively. FL-POLMPSO and FL-TOLMPSO
proposed in [41] converges 10−3.5 and 10−5.5 MMSE at 150th
and 160th NoCs, respectively. And Proposed FLeCSI-POLMPSO
and FLeCSI- TOLMPSO gives 4.6017*10−6, 4.6016*10−6 MMSE
Figure 6 Proposed FLeCSI-based lookup diagram for medium CI &
medium SI.
Figure 7 Proposed Fuzzy Logic empowered Cognitive and Social
Information (FLeCSI)-based lookup diagram for high CI and high SI.
8M. Asadullah et al. / International Journal of Computational Intelligence Systems, in press
Table 3 Comparison with the state-of-the-art method w.r.t NoC vs
MMSE.
Method Number of
Cycles
Min Mean Square
Error (SNR = 25dB)
POMPSO [13,18] 160 10−3
TOMPSO [13,18] 180 10−5.4
FL-POLMPSO [18] 150 10−3.5
FL-TOLMPSO [18] 160 10−5.5
Proposed FLeCSI-POLMPSO 117 4.6017 × 10−6
Proposed FLeCSI- TOLMPSO 110 4.6016 × 10−6
NoC, Number ofCycles; MMSE, Minimum Mean Square Error; POMPSO, Partial Opposite
Mutant Particle Swarm Optimization; TOMPSO, Total Opposite Mutant Particle Swarm
Optimization; FLeCSI, Fuzzy Logic empowered Cognitive and Social Information.
Table 4 Comparison with the state-of-the-art method w.r.t NoC vs
MMCE &and miss rate.
Method Minimum Mean
Channel Error (SNR
= 25dB, NoC = 200)
Miss Rate/Bit Error
Rate (SNR = 25dB,
NoC = 180)
POMPSO [13,18] 10−2 1 × 10−3
TOMPSO [13,18] 10−4 3.205 × 10−4
FL-POLMPSO [18] 10−3 3.8656 × 10−6
FL-TOLMPSO [18] 10−4.2 3.3633 × 10−6
Proposed FLeCSI-
POLMPSO
3.1944 × 10−5.1 1.523 × 10−6.5
Proposed FLeCSI-
TOLMPSO
3.001 × 10−5.1 1.5017 × 10−6.6
NoC, Number ofCycles; MMSE, Minimum Mean Square Error; POMPSO, Partial Opposite
Mutant Particle Swarm Optimization; TOMPSO, Total Opposite Mutant Particle Swarm
Optimization; FLeCSI, Fuzzy Logic empowered Cognitive and Social Information.
at 117th, 110th NoCs, respectively. It clearly observed Proposed
FLeCSI based on both variations give attractive results as compared
to previously published approaches in [13,18].
Table 4shows the performance of proposed FLeCSI based on
both variants in terms of MMCE and Mis Rate. The SNR is
fixed to 25 dB for both parameters (MMCE and Mis Rate).
The NoCs is also fixed to 200 for MMCE and 180 for Miss
Rate. It observed that Conventional POMPSO and Conven-
tional TOMPSO proposed in [13,18] gives 10−2, 10−4 MMCE and
10−3, 3.205*10−4 Miss Rate, respectively. FL-POLMPSO and FL-
TOLMPSO proposed in [35] converges 10−3, 10−4.2 MMCE and
3.8656*10−6, 3.3633*10−6 Miss Rate, respectively. And Proposed
FLeCSI-POLMPSO and FLeCSI-TOLMPSO gives 3.1944*10−5.1,
3.001*10−5.1 MMCE and 1.523*10−6.5, 1.5017*10−6.6 Miss Rate,
respectively. It clearly observed Proposed FLeCSI based both vari-
ations give attractive results as compared to previously published
approaches in [13,18] in terms of MMCE and Miss Rate.
5. CONCLUSION
Multiple Input Multiple Output (MIMO) technology is used for
the wireless communication systems, when it comes to the capac-
ity development of the channel. As the dynamic environment and
growing number of users, data rate demand increased in grow-
ing modern wireless communication systems. MUD with artifi-
cial intelligence plays a vital role to enhance network capacity to
meet the demands of future networks with an increased number
of users and multimedia services. The proposed FLeCSI-based
MIMO receiver system is fast enough to fulfill the needs of a future
generation growing networks. FLeCSI provides more attractive
results as compared to other suboptimal-based previous techniques
It is also observed that the proposed FLeCSI-based receiver system
minimizes BER and MMSE as compared to other techniques like
FL-OMPSO, FL-POLMPSO, FL-TOLMPSO, TOMPSO, etc. dis-
cussed previously.
6. FUTURE WORK
In this article, the Selective fading channel is ignored. Typ2
fuzzy may be used for updating the Cognition and Social infor-
mation of swarm intelligence. Deep machine learning may be
explored on CE along MUD for better results. The application
of this proposed FLeCSI can also be applied to medical imaging,
channel equalization, spectrum sensing in cognitive radio, etc. The
complexity comparison of proposed FLeCSI and their solution can
be another point of research.
DATA AVAILABILITY
The simulation files to support the findings of this study are avail-
able from the corresponding author upon request.
CONFLICT OF INTEREST
The authors declare that there are no conflicts of interest regarding
the publication of this article.
REFERENCES
[1] T. Abdelkader, K. Mokhtar, O. Abdelaziz, et al., New space-time
coding for joint blind channel estimation and data detection
through time-varying MIMO channels, Int. J. C omput. Sci. Issues.
11 (2014), 33–38.
[2] M. Abuthinien, S. Chen, A. Wolfgang, L. Hanzo, et al., Joint maxi-
mum likelihood channel estimation and data detection for MIMO
systems, in IEEE International Conference on Communications
ICC’07, Glasgow, UK, 2007, pp. 5354–5358.
[3] A. Mezghani, A.L.S. Hurst, Blind estimation of sparse multi-user
massive MIMO channels, in 21st International ITG Workshop on
Smart Antennas, Proceedings of VDE, 2016, vol. 2, pp. 2–6. http://
arxiv.org/abs/1612.00131
[4] M.N. Seyman, N. Taspinar, Symbol detection using the differen-
tial evolution algorithm in MIMO-OFDM systems, Turk. J. Electr.
Eng. Comput. Sci. 21 (2013), 373–380.
[5] M. Zubair, M.A. Choudhry, A. Naveed, I.M. Qureshi, et al., Joint
channel and data estimation using particle swarm optimization,
IEICE Trans. Commun. 91 (2008), 3033–3036.
[6] M.A. Khan, M. Umair, M.A.S. Choudhr y, et al., GA based adaptive
receiver for MC-CDMA system, Turk. J. Electr. Eng. Comput. Sci.
23 (2015), 2267–2277.
[7] M.A. Khan, M. Umair, M.A.S. Choudhry, et al., Island differen-
tial evolution based adaptive receiver for MC-CDMA system, in
IEEE International Conference on Information and Communica-
tion Technologies (ICICT), Karachi, Pakistan, 2015, pp. 1–6.
[8] M.A. Khan, A. Nasir, M. Umair, S. Abbas, et al., Time complex-
ity of TOMPSO algorithm, in 16th International Conference on
Statistical Sciences: Advances in Statistics and Data Management:
It’s Role in National Growth and Socio-Economic Developments,
2018, vol. 32, pp. 89–100.
M. Asadullah et al. / International Journal of Computational Intelligence Systems, in press 9
[9] P.W. Raut, S.L. Badjate, MIMO-future wireless communication,
Int. J. Innov. Technol. Explor. Eng. 2 (2013), 102–106.
[10] M. Biguesh, A.B. Gershman, Training-based MIMO channel esti-
mation: a study of estimator tradeoffs and optimal training sig-
nals, IEEE Trans. Signal Process. 54 (2006), 884–893.
[11] P. Chen, H. Kobayashi, Maximum likelihood channel estimation
and signal detection for OFDM systems, in IEEE International
Conference on Communication, 2002, vol. 3, pp. 1640–1645.
[12] C. Pirak, Z.J. Wang, K.R. Liu, S. Jitapunkul, et al., Optimum power
allocation for maximum-likelihood channel estimation in space-
time coded MIMO systems, in 2006 IEEE International Confer-
ence on Acoustics, Speech and Signal Processing, ICASSP 2006
Proceedings, Toulouse, France, 2006, vol. 4, pp. 4–6.
[13] M.A. Khan, Multiuser Detection Using Computational Intelli-
gence in Multi-Carrier Communication Systems, ISRA Univer-
sity, Islamabad Campus, Pakistan, 2016.
[14] M. Umair, M.A. Khan, M.A.S. Choudhry, et al., Island genetic
algorithm based MUD for MC-CDMA system, in 6th IEEE Inter-
national Conference on Information and Communication Tech-
nologies, Electronic, Karachi, Pakistan, 2015, pp. 7–12.
[15] N. Ali, M.A. Khan, M. Adeel, M. Amir, et al., Genetic algorithm
based adaptive receiver for MC-CDMA system with variation in
mutation operator, Int. J. Comput. Sci. Inf. Secur. USA. 14 (2016),
215–222.
[16] M. Umair, M.A. Khan, M.A.S. Choudhry, et al., Piranha fish opti-
mization for multi-user detection in OFDMA system, Int. J. Adv.
Appl. Sci. 3 (2016), 35–40.
[17] M. Umair, M.A. Khan, M.A.S. Choudhry, et al., GA backing
to STBC based MC-CDMA systems, in 4th IEEE International
Conference on Intelligent Systems, Modelling, and Simulation,
Bangkok, Thailand, 2013, pp. 503–506.
[18] M. AsadUllah, M.A. Khan, S. Abbas, A. Athar, S.S. Raza,
G. Ahmad, et al., Blind channel and data estimation using fuzzy
logic-empowered opposite learning-based mutant particle swarm
optimization, Comput. Intell. Neurosci. 2018 (2018), 1–12.
[19] R.C. Eberhart, J. Kennedy, A new optimizer using particle swarm
theory, in Proceedings of the Sixth International Symposium
on Micro Machine and Human Science, Nagoya, Japan, 1995,
pp. 39–43.
[20] J. Kennedy, R.C. Eberhart, Particle swarm optimization, in Pro-
ceedings of the IEEE International Conference on Neural Net-
works, Piscataway, NJ, USA, 1995, vol. 4, pp. 1942–1948.
[21] J. Kennedy, R.C. Eberhart, Swarm Intelligence, MorganKaufmann
Publishers, Massachusetts, USA, 2001.
[22] H.Z. Wei, C. Yang, L. Gao, X. Yao, R.C. Eberhart, Y. Shi, et al.,
Extracting rules from fuzzy neural network by particle swarm
optimization, in Proceedings of IEEE International Conference
on Evolutionary Computation, Anchorage, AK, USA, 1998,
pp. 74–77.
[23] A. Sahu, S.K. Panigrahi, S. Pattnaik, et al., Fast convergence par-
ticle swarm optimization for functions optimization, Procedia
Technol. 4 (2012), 319–324.
[24] M. Clerc, J. Kennedy, The particle swarm explosion, stability and
convergence in a multidimensional complex space, IEEE Trans.
Evol. Comput. 6 (2002), 58–73.
[25] B. Liu, L. Wang, Y.H. Jin, F. Tang, D.X. Huang, et al., Improved
particle swarm optimization combined with chaos,C haos Soliton.
Fract. 25 (2005), 1261–71.
[26] R.C. Eberhart, Y. Shi, Comparison between genetic algorithms
and particle swarm optimization, in Evolutionary Program-
ming VII: Proceedings of the Seventh Annual Conference
on Evolutionary Programming, San Diego, CA, USA, 1998,
pp. 611–616.
[27] X.H. Shi, Y.C. Liang, H.P. Lee, C. Lu, L.M. Wang, et al., An
improved GA and a novel PSO-GA-based hybrid algorithm, Inf.
Process. Lett. 93 (2005), 255–261.
[28] P.N. Suganthan, Particle swarm optimizer with neighborhood
operator, in Proceedings of the 1999 Congress on Evolu-
tionary Computation, CEC’99, Washington, DC, USA, 1999,
pp. 1958–1961.
[29] A. Nickabadi, M.M. Ebadzadeh, R. Safabakhsh, et al., A novel par-
ticle swarm optimization algorithm with adaptive inertia weight,
Appl. Soft Comput. 11 (2011), 3658–3670.
[30] A. Chatterjee, P. Siarry, Nonlinear inertia weight variation for
dynamic adaptation in particle swarm optimization, Comput.
Oper. Res. 33 (2006), 859–871.
[31] A. Alfi, PSO with adaptive mutation and inertia weight and its
application in parameter estimation of dynamic systems, Acta
Automat. Sin. 37 (2011), 541–549.
[32] H. Modares, A. Alfi, M.M. Fateh, et al., Parameter identification
of chaotic dynamic systems through an improved particle swarm
optimization, Expert Syst. Appl. 37 (2010), 3714–3720.
[33] P. Civicioglu, Backtracking search optimization algorithm for
numerical optimization problems, Appl. Math. Comput. 219
(2013), 8121–8144.
[34] Y. Shi, R.C. Eberhart, An empirical study of particle swarm opti-
mization, in Proceedings of the IEEE Congress on Evolutionary
Computation, Piscataway, NJ, USA, 1999, pp. 1945–1960.
[35] F.V. Bergh, A.P. Engelbrech, A study of particle swarm optimiza-
tion particle trajectories, Inf. Sci. 176 (2006), 937–971.
[36] D. Tripathi, U. Jena, Cognitive and social information based PSO,
Int. J. Eng. Sci. Technol. 8 (2016), 64–67.
[37] R. Brits, A. Engelbrecht, F.V. Bergh, et al., Locating multiple
optima using particle swarm optimization, Appl. Math. Comput.
189 (2007), 1859–1883.
